Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!dalcs!thompson
From: thompson@dalcs.UUCP (Michael A. Thompson)
Newsgroups: comp.unix.questions
Subject: Re: Wierd Compilers
Message-ID: <2678@dalcs.UUCP>
Date: Tue, 3-Nov-87 12:29:01 EST
Article-I.D.: dalcs.2678
Posted: Tue Nov  3 12:29:01 1987
Date-Received: Fri, 6-Nov-87 22:36:41 EST
References: <367@white.gcm>
Reply-To: thompson@dalcs.UUCP (Michael A. Thompson)
Organization: Math, Stats & CS, Dalhousie University, Halifax, NS, Canada
Lines: 48
Keywords: Sun

In article <367@white.gcm> dc@gcm (Dave Caswell) writes:
>
>Does anyone have a guess why program one runs in half the time of program two?
>BTW this is a SUN 3.

	The same occures on our VAX 11/785 running BSD4.3:
Program 1		0.7u 0.0s 0:00 81% 1+5k 0+0io 1pf+0w
Program 1 w/-O		0.7u 0.0s 0:00 97% 1+5k 0+1io 1pf+0w
Program 2		1.1u 0.0s 0:01 97% 1+5k 0+1io 1pf+0w
Program 2 w/-O		0.9u 0.0s 0:01 71% 1+5k 0+1io 1pf+0w

	Here is the assembler output from our C compiler:
Program 1:                      Program 2:
	.			        .
	.			        .
	.			        .
L16:				L16:
	decl	-4(fp)		        movl    -4(fp),r0
	jeql	L17		        decl    -4(fp)
	jbr 	L16		        tstl    r0
L17:				        jeql    L17
	.			        jbr     L16
	.			L17:
	.			        .
				        .
				        .

Program 1 w/-O:                 Program 2 w/-O:
	.			        .
	.			        .
	.			        .
L16:decl	-4(fp)		L16:movl        -4(fp),r0
jneq	L16			decl    -4(fp)
	.			tstl    r0
	.			jneq    L16
	.			        .
				        .
				        .

	Without trying to go into why the compiler constructs it's
    loops this way, it is pretty obvious that program 2 is taking
    longer because it has more instructions to execute each time
    through the loop.
-- 
Michael A. Thompson, Dept. Math, Stats, & C.S., Dalhousie U., Halifax, N.S.
thompson@dalcs.uucp	From Bitnet or Uucp
thompson@cs.dal.cdn	From Bitnet or Cdn
thompson%dalcs.uucp@uunet.uu.net From Arpa